Skip to main content
skill CX Codex

sales-execution — Claude Code skill: what it does, how to install

Logs calls, notes, meetings, and tasks in HubSpot CRM via the CLI. Links every entry to contacts and deals so records appear in the UI.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add HubSpot/agent-cli-skills --skill sales-execution -a claude-code
Open on GitHub

What sales-execution does

This skill guides coding agents through recording sales actions with the HubSpot CLI. It handles calls, notes, meetings, and tasks. Timestamps on creation require 13-digit Unix milliseconds. For tasks, hs_timestamp acts as the due date rather than the creation time.

Creating an object alone leaves it invisible in the CRM interface. Every call to hubspot objects create must immediately precede hubspot associations create targeting a deal or contact. For bulk follow-ups, the skill pipes JSONL streams into batch CLI commands, mapping IDs line-by-line via paste and jq.

Enrolling contacts into Sales Hub sequences is unsupported. Sequence endpoints in the CLI are read-only, so automated outreach requires scheduling direct tasks instead.

What it helps with

  • Log outbound calls and discovery notes directly to HubSpot deals
  • Schedule high-priority follow-up tasks with future Unix millisecond deadlines
  • Attach meeting records to contacts and deals simultaneously
  • Generate follow-up tasks for all deals currently in a specific pipeline stage

Example prompts for the agent

  • Log a discovery call for deal 456 and contact 149 with completed status
  • Create a follow-up call task for deal 456 due in 7 days
  • Log a note on deal 456 saying the proposal was sent

How to install sales-execution

1Skills CLI (Vercel)

npx skills add HubSpot/agent-cli-skills --skill sales-execution -a claude-code

2Manually: put the files in place

~/.claude/skills/sales-execution/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/sales-execution/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about sales-execution

How to install sales-execution in Claude Code?

Run npx skills add HubSpot/agent-cli-skills --skill sales-execution -a claude-code. The skill definition will be placed in ~/.claude/skills/sales-execution/SKILL.md.

What does the sales-execution skill do?

It provides procedural rules and CLI patterns for creating HubSpot engagements and immediately associating them with CRM records.

Can this skill enroll contacts into HubSpot sequences?

No, because HubSpot CLI sequence commands operate strictly under read-only scopes. Reps must schedule outreach tasks or log activities directly.

Source text

SKILL.md · Apache-2.0
---
name: sales-execution
description: Log sales activities — calls, notes, meetings, tasks — against contacts and deals, with the mandatory create-then-associate step that makes them visible in the CRM.
triggers:
  - "log a call"
  - "log a note"
  - "log a meeting"
  - "create task"
  - "follow-up task"
  - "log activity"
  - "log engagement"
---

## Resources

| File | When to use |
|---|---|
| `resources/activity-properties-reference.md` | Property names and enum values for calls/notes/meetings/tasks. Keep open while writing `objects create` — for live enum values run `hubspot properties options-list --type calls <property>`, and for call outcomes `hubspot activities calls dispositions list`; the reference is the shortcut. |

Read `bulk-operations/SKILL.md` first — this skill assumes its batching, pipe, and dry-run patterns.

## The two non-obvious rules

**1. Activities are invisible until associated.** `hubspot objects create --type calls ...` alone produces a record nobody can see in the CRM UI. Always follow with `hubspot associations create --from calls:<id> --to contacts:<id>` (and the deal, if relevant) before stopping.

**2. Timestamps differ between write and read.**

| Path | Field | Format |
|---|---|---|
| `objects create --property hs_timestamp=...` | `hs_timestamp` | **Unix ms** (13 digits) |
| `objects get --type calls <id>` returns | `properties.hs_timestamp` | Unix ms (string) |
| `activities list --contact <id>` returns | `timestamp` (flat, top-level) | **ISO 8601** (e.g. `2024-01-15T10:00:00Z`) |

Current Unix ms: `$(date +%s)000` (macOS) or `$(date +%s%3N)` (Linux). `activities list` rows are `{"id","type","timestamp","lastModified","title","body","status","owner_id"}` — the cross-type timeline read shape, no raw property names.

## Create + associate, by type

```bash
# CALL
call_id=$(hubspot objects create --type calls \
  --property hs_call_title="Discovery call" \
  --property hs_call_body="Confirmed $50K budget, Q2 timeline." \
  --property hs_call_direction=OUTBOUND \
  --property hs_call_status=COMPLETED \
  --property hs_call_duration=1800000 \
  --property hs_timestamp=$(date +%s)000 \
  --format json | jq -r '.id')
hubspot associations create --from calls:$call_id --to contacts:149
hubspot associations create --from calls:$call_id --to deals:456

# NOTE
note_id=$(hubspot objects create --type notes \
  --property hs_note_body="Sent proposal. Follow-up Friday." \
  --property hs_timestamp=$(date +%s)000 \
  --format json | jq -r '.id')
hubspot associations create --from notes:$note_id --to deals:456

# MEETING — start/end in Unix ms; reuse start as hs_timestamp
start=$(date +%s)000; end=$(( ${start%000} + 3600 ))000
meeting_id=$(hubspot objects create --type meetings \
  --property hs_meeting_title="Demo — Acme" --property hs_meeting_outcome=COMPLETED \
  --property hs_meeting_start_time=$start --property hs_meeting_end_time=$end \
  --property hs_timestamp=$start --format json | jq -r '.id')
hubspot associations create --from meetings:$meeting_id --to contacts:149

# TASK — hs_timestamp is the DUE DATE, not creation time
due=$(( $(date -v+7d +%s) * 1000 ))   # macOS; Linux: date -d '7 days' +%s
task_id=$(hubspot objects create --type tasks \
  --property hs_task_subject="Confirm proposal received" \
  --property hs_task_priority=HIGH \
  --property hs_task_status=NOT_STARTED \
  --property hs_task_type=CALL \
  --property hs_timestamp=$due \
  --format json | jq -r '.id')
hubspot associations create --from tasks:$task_id --to deals:456
```

## Open tasks for a contact — two CLI calls, no xargs

`associations list` emits `{"id": …}` per row (plus label/associationType fields); `objects get` reads from stdin in one batch call (see `bulk-operations/SKILL.md` "Read in batch").

```bash
hubspot associations list --from contacts:149 --to tasks \
| hubspot objects get --type tasks \
    --properties hs_task_subject,hs_task_status,hs_task_priority,hs_timestamp \
| jq -c 'select(.properties.hs_task_status != "COMPLETED")'
```

## Open tasks portal-wide — exclude statuses server-side with `@!=`

`@!=` is NOT_IN: one condition for the whole excluded list. Do **not** spend one `--filter` flag per excluded status — the search API caps a request at 5 filter groups, 6 conditions per group, and 18 conditions total.

`@!=` matches only records that *have* the property, so replicating a real HubSpot task view's "status is none of … or is empty" needs two groups (OR'd), with the shared conditions repeated in each:

```bash
hubspot objects search --type tasks \
  --filter "hubspot_owner_id=123 AND hs_task_status@!=COMPLETED,DEFERRED" \
  --filter "hubspot_owner_id=123 AND !hs_task_status" \
  --properties hs_task_subject,hs_task_status,hs_task_priority,hs_timestamp
```

Inverse (only these statuses) is `@=` (IN): `--filter "hs_task_status@=NOT_STARTED,IN_PROGRESS,WAITING"`.

## Bulk: follow-up task per deal in a stage

The deal ID and the task ID must travel together. Persist the deal payload to a file, create tasks (output order matches input order — see bulk-operations), then zip the two ID lists line-by-line and stream association pairs in one call.

```bash
due=$(( $(date -v+7d +%s) * 1000 ))

# 1. Per-deal payload, deal_id retained alongside the create payload.
hubspot objects search --type deals --filter "dealstage=appointmentscheduled" \
  --properties dealname \
| jq -c --argjson due "$due" '{deal_id: .id, payload: {properties: {
    hs_task_subject: ("Follow up: " + .properties.dealname),
    hs_task_priority: "HIGH", hs_task_status: "NOT_STARTED", hs_task_type: "CALL",
    hs_timestamp: ($due|tostring)
  }}}' > /tmp/deal_tasks.jsonl

# 2. Create tasks; one CLI call for the whole batch.
jq -c '.payload' /tmp/deal_tasks.jsonl \
| hubspot objects create --type tasks > /tmp/created_tasks.jsonl

# 3. Zip and stream association pairs through stdin.
paste \
  <(jq -r '.deal_id' /tmp/deal_tasks.jsonl) \
  <(jq -r '.id'      /tmp/created_tasks.jsonl) \
| jq -Rc 'split("\t") | {from:("tasks:"+.[1]), to:("deals:"+.[0])}' \
| hubspot associations create
```

These are `objects create` calls (Additive) — creates are never digest-gated, so `--dry-run` is a plain preview and you execute by dropping it (no digest/confirm). The digest flow only applies to the irreversible writes (`update`/`upsert`/`delete`/`merge`) in `bulk-operations/SKILL.md`.

## Known constraints

Activities must be associated immediately or they're invisible in the CRM UI. For enum option values on activity properties, run `hubspot properties options-list --type calls <property>` (and for call outcomes, `hubspot activities calls dispositions list`); the reference is the shortcut. Sales Hub sequences are read-only in the CLI (`hubspot sequences list` / `get` / `enrollments`, `automation.sequences.read` scope) — the CLI cannot enroll a contact in a sequence, so drive outreach via tasks/activities above. This surface grows; recheck `hubspot --help` / `CHANGELOG.md` before assuming an API is missing.

© HubSpot · Apache-2.0 · HubSpot/agent-cli-skills

Similar skills

All skills for Notes, tasks and projects →

From the same set15

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor